Charles Ryan: We are at the Prostate Cancer Foundation's 25th anniversary retreat, and I'm delighted to be joined by Andy Armstrong, a colleague and friend. I've known you for a long time. We're going to talk about some of your ongoing work. Andy is recently promoted to Full Professor of Medicine at Duke ...

Andrew Armstrong: Thank you, Chuck.

Charles Ryan: ... Medical School. Congratulations on that big accomplishment. And you are an Associate Professor in Pharmacology and Cancer Biology and Associate Professor in Surgery, and you're a member of the Duke Cancer Institute, so lots of titles. But anyway, thank you for joining us, and we're going to talk today about your study, the PROPHECY study. So the PROPHECY is an acronym. What's the acronym, and what's the study?

Andrew Armstrong: So the PROPHECY was a PCF Challenge Award co-funded by Movember. So speaking of which, the month is about to start, a great fundraising month for prostate cancer health, so I'd encourage everybody to grow a mustache and support the cause.

Charles Ryan: Yeah, I'm itching under my nose already-

Andrew Armstrong: Absolutely.

Charles Ryan: -in anticipation of growing a mustache.

Andrew Armstrong: So PROPHECY... One of the ways that we like to come up with acronyms is, we think of what it will need to stand for and then make it fit. So predictive biomarkers for prostate cancer, being able to see the future was the acronym basically. And the PROPHECY study is a predictive biomarker study of a variety of liquid biopsies for men with castrate-resistant prostate cancer who are undergoing therapy with a range of therapies, abiraterone and enzalutamide primarily. That was the primary aim of the grant. And then taxane chemotherapy is a secondary aim.

Charles Ryan: So are you screening for predictive biomarkers, or are you testing known biomarkers to see if they're predictive?

Andrew Armstrong: Yeah, both actually. So men were very altruistic in this study. It was 120 men. They consented to up to 90 mls of blood that was drawn. The blood was screened for a number of already validated biomarkers, but ones that hadn't yet been prospectively tested, such as the Hopkins AR-V7 test, the Epic Sciences AR-V7 nuclear test. Those have been validated in single-center studies at Hopkins and at Memorial Sloan Kettering but have not been prospectively validated multi-center kind of blinded studies. And that was one of the primary objectives of PROPHECY, was to take two AR-V7 tests, compare them against each other, compare them in a multi-variable way to see how well do they predict the future, how well do they predict response. Progression-free survival was the primary objective, and then overall survival.

Charles Ryan: 90 mls of blood. That's-

Andrew Armstrong: It's quite a lot.

Charles Ryan: Quite a bit.

Andrew Armstrong: Men make that much blood up in about a week, so it's pretty fast recovery. Most men contribute about 50 mls, but a subset of these men had what's called the high-throughput analyses, so they had their circulating tumor cells analyzed for whole-genome sequencing, whole-exome sequencing, RNA sequencing, kind of the works, cell-free DNA. So we have that kind of compendium of molecular tools to look at to explain, well, of the AR-V7 negative men, some men still had a lot of resistance, developed resistance quickly. What other predictors can we find in their genome using just a simple liquid biopsy to predict the future?

Charles Ryan: Are you going to be able to detect... Of course, detecting mutations is reasonable in circulating tumor DNA. We can do a lot with circulating tumor cells. There are some circumstances where there is a loss that occurs, for example, RB and other things, and these are critically important biological pathways that, if we have the tissue we can do the test, but it's not quite clear that we can do this on blood.

Andrew Armstrong: It's a great question. When we enrich the samples for CTCs, we're actually able to see losses to a greater degree than the same patient's blood at the same time point using cell-free DNA.

Charles Ryan: Interesting. Okay.

Andrew Armstrong: If you think of the cell-free DNA, it's like a smoothie analogy where there's a lot of normal DNA and you're trying to sample a loss. It's really challenging. But if you enrich the sample for actually tumor content, much like a biopsy is enriched, you can see those things.

Charles Ryan: And how do you enrich the sample for tumor content?

Andrew Armstrong: Well, to get to the circulating tumor cells, there's a number of methods. Epic does it by picking the cells off the slide. We've done it in our approach by using flow cytometry. So depleting the normal leukocytes, depleting the red cells, and then enriching for epithelial characteristics allows us to enrich, not purify but enrich, and get rid of the noise.

Charles Ryan: So you have a sample that is essentially what you believe to be mostly circulating tumor cells.

Andrew Armstrong: Mostly tumors, yeah.

Charles Ryan: And so, therefore, you can look along the normal backdrop, so to speak, to see if there's RB loss or other things like-

Andrew Armstrong: Yeah, in the same patient we draw their germline DNA, so we actually cross-hybridize the genomes against the germline genome so we can see what's lost in their tumor sample versus their non-tumor sample.

Charles Ryan: Okay. You mentioned AR-V7 earlier. What's your perspective on where we are with that story right now?

Andrew Armstrong: So the single-center experiences have been pretty interesting. Emmanuel Antonarakis published in the New England Journal first on the predictive value for abi-enza resistance was almost an all-or-nothing response. And I think, based on 30 patients with abi, 30 with enza, that was a very profound signal.

His follow-up study for the Hopkins experience with over 200 patients has really showed that there are three groups. There are patients without CTCs, and then there are patients with CTCs. And then of the CTC positive groups, there's those that have V7 and those who don't. The group of patients that do the best are those without CTCs. They have a good prognosis. The test is considered uninformative because there's no tumor, no tumor cells to measure. The patients with the worst prognosis are those with CTCs and V7. They have a lousy outcome with abi and enza. They actually have a poor prognosis for basically all known therapies, but they seem to have a benefit from taxanes. These are rapidly proliferating cells. And how our chairs work with Epic has mirrored that. The Epic CTCs as well seem to predict for resistance to AR-directed therapies where the lagging and binding domain's missing, so biologically it makes sense. Taxanes seem to be agnostic of that, and they destroy rapidly proliferating cells.

So the idea would be a companion diagnostic kind of test for patients who have failed abi or enza. A physician is trying to decide on whether they should get the other one or a taxane or some other experimental trial. And if you can have a simple blood test that would say the chances of the drug working is close to zero, it may be worth it. The drugs are expensive, and they have side effects, and most importantly they waste a patient's time if they're not going to work.

Charles Ryan: So tell us a little bit, and I know this is really complicated, about the credentialing of these biomarkers for treatment selection. So you mentioned if you could identify AR-V7 you as the oncologist might say, "I think I'll treat this patient with the taxane as opposed to enzalutamide or something like that." but if you're the FDA and/or you are trying to develop a new drug and you want to use that bio-marker reproducibly and make it work for everybody across the world, how do you do that? And how much of that type of thinking enters into what you're doing?

Andrew Armstrong: Sure. So biomarker development is as complicated if not more so than drug development.

Charles Ryan: Probably more, yeah.

Andrew Armstrong: And it goes through analytic validation, which is kind of like the preclinical drug testing. You want to make sure you know what you're measuring, that it's reproducible, that you've got an assay that will survive in blood, that will be reporting on what it's intending, that it has good sensitivity, that it's not producing false positives and things like that.

Both Epic and Hopkins have put a lot of time into that analytic validation. The Hopkins assay is an RT-PCR-based assay, so they've come up with an RT-PCR-based number of cycles of amplification above which is called a positive test. And they use controls, the presence of full-length AR, PSA, EpCAM. It was all in there to assure that we have CTCs in the sample. The Epic assay is a little more complicated. It's more like a blood smear. So it's an automated software program that can read nuclear AR-V7. So if it's in the cytoplasm but not the nucleus, it's not thought to be active, and that's not as predictive as when you see it actually doing its business in the nucleus. And they've used an antibody that seems to have clinical validity.

And so clinical validation goes through Phase I, II, III. Ideally controlled studies will be done. Right now, I would say it's gone through Phase II for single-center studies, just like a Phase II drug trial. PROPHECY I would consider as more like a controlled study where you have prospective validation. The definitions of each biomarker are locked in ahead of time. They're in the protocol. So unlike the previous work where they kind of built the biomarker and then looked at it in retrospect, PROPHECY had it all locked in ahead of time.

And then we did a multi-center study. So if you're doing a study, you don't want it to just be in your lab. You want it to be a sample that will survive shipping across the country, that will... Even if it's overnight, that it shows that it's reproducible in a clinical context. We had five centers and five central labs, so a massive effort for our research coordinators to get these blood samples all over the country.

Charles Ryan: Great. So where do you go? You said PROPHECY is about seeing the future. PROPHECY is enrolling now.

Andrew Armstrong: No, it finished.

Charles Ryan: It finished.

Andrew Armstrong: Yeah, so-

Charles Ryan: And where are you going with it?

Andrew Armstrong: So we've submitted it for publication, presented it at ASCO, and I'll be presenting it at this meeting. And we basically showed that both AR-V7 assays kind of passed their hurdle. In the univariate analysis, they were associated with a low chance of PSA response, confirmed PSA decline. The progression-free survival in patients that had a positive test was like three months. It was twice that if you had no AR-V7 with both assays, and the overall survival was almost three to four times longer in the V7-negative patients versus the positive patients. And more importantly-

Charles Ryan: Three to four times longer?

Andrew Armstrong: Yes, so a hazard ratio of about four.

Charles Ryan: That's-

Andrew Armstrong: Big.

Charles Ryan: -pretty amazing. Yeah.

Andrew Armstrong: Yeah. The biggest question for AR-V7 is, is it just reporting on disease burden? Okay? So in the original studies, the adjustments were not made for the number of CTCs or things like the Halabi risk group. So Susan Halabi is our statistician on the study, so we've now included adjustments for the Halabi risk group, and then the CELLSEARCH® CTC enumeration, which was performed in a CLIA-certified lab. So the good news is the multi-variable adjustment was also significant for both endpoints, and the estimates of hazard ratios didn't change much when you adjust for the number of CTCs, meaning disease burden, and other poor prognostic factors. So it's not just disease burden. It is reporting on some aspect of biology in a subset of these men.

Charles Ryan: Got it.

Andrew Armstrong: Another main message is that there's a lot of men who have resistance but don't have AR-V7. So I think that's kind of lost in a lot of the literature, that AR-V7 is going to solve everything, but it's clearly not. But for those men who are positive, it's an important signal. For those men who are negative, there's still other mechanisms of resistance, and I think that's where the PROPHECY study is also going to contribute. We can look for genomic amplifications, deletions that are associated with neuroendocrine disease. And MIC, for example, RB loss, PTEN, P53. Those are pretty aggressive genotypes, but they're not necessarily associate AR-V7.

Charles Ryan: Right. I think we're going to get to the point where prostate cancer becomes a little bit like we see now non-small-cell lung cancer.

Andrew Armstrong: Exactly.

Charles Ryan: Where we have this pie chart-

Andrew Armstrong: A pie chart.

Charles Ryan: -of mutations and alterations and therapies, and I think we're not going to get there by doing... I think the biopsies are just harder to do. That's right. So we have to come up with a liquid process for this. You're out there at the forefront of doing this. Congratulations on your promotion to Full Professor.

Andrew Armstrong: Thank you.

Charles Ryan: Publication of the PROPHECY, and really in leading the field and being part of the leadership in this field of moving this liquid biopsy piece forward. I think it stands to benefit a lot of us physicians for our decision-making, but also obviously patients for selecting the right therapy for them at the right time.

Andrew Armstrong: Thank you, Jack.

Charles Ryan: So congratulations and thanks for joining us.

Andrew Armstrong: All right. Thank you.